C++ ConstantInt类代码示例

int TargetTransformInfo::getInstructionThroughput(const Instruction *I) const {
  switch (I->getOpcode()) {
  case Instruction::GetElementPtr:
    return getUserCost(I);

  case Instruction::Ret:
  case Instruction::PHI:
  case Instruction::Br: {
    return getCFInstrCost(I->getOpcode());
  }
  case Instruction::Add:
  case Instruction::FAdd:
  case Instruction::Sub:
  case Instruction::FSub:
  case Instruction::Mul:
  case Instruction::FMul:
  case Instruction::UDiv:
  case Instruction::SDiv:
  case Instruction::FDiv:
  case Instruction::URem:
  case Instruction::SRem:
  case Instruction::FRem:
  case Instruction::Shl:
  case Instruction::LShr:
  case Instruction::AShr:
  case Instruction::And:
  case Instruction::Or:
  case Instruction::Xor: {
    TargetTransformInfo::OperandValueKind Op1VK =
      getOperandInfo(I->getOperand(0));
    TargetTransformInfo::OperandValueKind Op2VK =
      getOperandInfo(I->getOperand(1));
    SmallVector<const Value*, 2> Operands(I->operand_values());
    return getArithmeticInstrCost(I->getOpcode(), I->getType(), Op1VK,
                                       Op2VK, TargetTransformInfo::OP_None,
                                       TargetTransformInfo::OP_None,
                                       Operands);
  }
  case Instruction::Select: {
    const SelectInst *SI = cast<SelectInst>(I);
    Type *CondTy = SI->getCondition()->getType();
    return getCmpSelInstrCost(I->getOpcode(), I->getType(), CondTy, I);
  }
  case Instruction::ICmp:
  case Instruction::FCmp: {
    Type *ValTy = I->getOperand(0)->getType();
    return getCmpSelInstrCost(I->getOpcode(), ValTy, I->getType(), I);
  }
  case Instruction::Store: {
    const StoreInst *SI = cast<StoreInst>(I);
    Type *ValTy = SI->getValueOperand()->getType();
    return getMemoryOpCost(I->getOpcode(), ValTy,
                                SI->getAlignment(),
                                SI->getPointerAddressSpace(), I);
  }
  case Instruction::Load: {
    const LoadInst *LI = cast<LoadInst>(I);
    return getMemoryOpCost(I->getOpcode(), I->getType(),
                                LI->getAlignment(),
                                LI->getPointerAddressSpace(), I);
  }
  case Instruction::ZExt:
  case Instruction::SExt:
  case Instruction::FPToUI:
  case Instruction::FPToSI:
  case Instruction::FPExt:
  case Instruction::PtrToInt:
  case Instruction::IntToPtr:
  case Instruction::SIToFP:
  case Instruction::UIToFP:
  case Instruction::Trunc:
  case Instruction::FPTrunc:
  case Instruction::BitCast:
  case Instruction::AddrSpaceCast: {
    Type *SrcTy = I->getOperand(0)->getType();
    return getCastInstrCost(I->getOpcode(), I->getType(), SrcTy, I);
  }
  case Instruction::ExtractElement: {
    const ExtractElementInst * EEI = cast<ExtractElementInst>(I);
    ConstantInt *CI = dyn_cast<ConstantInt>(I->getOperand(1));
    unsigned Idx = -1;
    if (CI)
      Idx = CI->getZExtValue();

    // Try to match a reduction sequence (series of shufflevector and vector
    // adds followed by a extractelement).
    unsigned ReduxOpCode;
    Type *ReduxType;

    switch (matchVectorSplittingReduction(EEI, ReduxOpCode, ReduxType)) {
    case RK_Arithmetic:
      return getArithmeticReductionCost(ReduxOpCode, ReduxType,
                                             /*IsPairwiseForm=*/false);
    case RK_MinMax:
      return getMinMaxReductionCost(
          ReduxType, CmpInst::makeCmpResultType(ReduxType),
          /*IsPairwiseForm=*/false, /*IsUnsigned=*/false);
    case RK_UnsignedMinMax:
      return getMinMaxReductionCost(
          ReduxType, CmpInst::makeCmpResultType(ReduxType),
//.........这里部分代码省略.........

unsigned CostModelAnalysis::getInstructionCost(const Instruction *I) const {
  if (!TTI)
    return -1;

  switch (I->getOpcode()) {
  case Instruction::GetElementPtr:{
    Type *ValTy = I->getOperand(0)->getType()->getPointerElementType();
    return TTI->getAddressComputationCost(ValTy);
  }

  case Instruction::Ret:
  case Instruction::PHI:
  case Instruction::Br: {
    return TTI->getCFInstrCost(I->getOpcode());
  }
  case Instruction::Add:
  case Instruction::FAdd:
  case Instruction::Sub:
  case Instruction::FSub:
  case Instruction::Mul:
  case Instruction::FMul:
  case Instruction::UDiv:
  case Instruction::SDiv:
  case Instruction::FDiv:
  case Instruction::URem:
  case Instruction::SRem:
  case Instruction::FRem:
  case Instruction::Shl:
  case Instruction::LShr:
  case Instruction::AShr:
  case Instruction::And:
  case Instruction::Or:
  case Instruction::Xor: {
    return TTI->getArithmeticInstrCost(I->getOpcode(), I->getType());
  }
  case Instruction::Select: {
    const SelectInst *SI = cast<SelectInst>(I);
    Type *CondTy = SI->getCondition()->getType();
    return TTI->getCmpSelInstrCost(I->getOpcode(), I->getType(), CondTy);
  }
  case Instruction::ICmp:
  case Instruction::FCmp: {
    Type *ValTy = I->getOperand(0)->getType();
    return TTI->getCmpSelInstrCost(I->getOpcode(), ValTy);
  }
  case Instruction::Store: {
    const StoreInst *SI = cast<StoreInst>(I);
    Type *ValTy = SI->getValueOperand()->getType();
    return TTI->getMemoryOpCost(I->getOpcode(), ValTy,
                                 SI->getAlignment(),
                                 SI->getPointerAddressSpace());
  }
  case Instruction::Load: {
    const LoadInst *LI = cast<LoadInst>(I);
    return TTI->getMemoryOpCost(I->getOpcode(), I->getType(),
                                 LI->getAlignment(),
                                 LI->getPointerAddressSpace());
  }
  case Instruction::ZExt:
  case Instruction::SExt:
  case Instruction::FPToUI:
  case Instruction::FPToSI:
  case Instruction::FPExt:
  case Instruction::PtrToInt:
  case Instruction::IntToPtr:
  case Instruction::SIToFP:
  case Instruction::UIToFP:
  case Instruction::Trunc:
  case Instruction::FPTrunc:
  case Instruction::BitCast: {
    Type *SrcTy = I->getOperand(0)->getType();
    return TTI->getCastInstrCost(I->getOpcode(), I->getType(), SrcTy);
  }
  case Instruction::ExtractElement: {
    const ExtractElementInst * EEI = cast<ExtractElementInst>(I);
    ConstantInt *CI = dyn_cast<ConstantInt>(I->getOperand(1));
    unsigned Idx = -1;
    if (CI)
      Idx = CI->getZExtValue();
    return TTI->getVectorInstrCost(I->getOpcode(),
                                   EEI->getOperand(0)->getType(), Idx);
  }
  case Instruction::InsertElement: {
      const InsertElementInst * IE = cast<InsertElementInst>(I);
      ConstantInt *CI = dyn_cast<ConstantInt>(IE->getOperand(2));
      unsigned Idx = -1;
      if (CI)
        Idx = CI->getZExtValue();
      return TTI->getVectorInstrCost(I->getOpcode(),
                                     IE->getType(), Idx);
    }
  case Instruction::ShuffleVector: {
    const ShuffleVectorInst *Shuffle = cast<ShuffleVectorInst>(I);
    Type *VecTypOp0 = Shuffle->getOperand(0)->getType();
    unsigned NumVecElems = VecTypOp0->getVectorNumElements();
    SmallVector<int, 16> Mask = Shuffle->getShuffleMask();

    if (NumVecElems == Mask.size() && isReverseVectorMask(Mask))
      return TTI->getShuffleCost(TargetTransformInfo::SK_Reverse, VecTypOp0, 0,
                                 0);
//.........这里部分代码省略.........

/// CloneBlock - The specified block is found to be reachable, clone it and
/// anything that it can reach.
void PruningFunctionCloner::CloneBlock(const BasicBlock *BB,
                                       std::vector<const BasicBlock*> &ToClone){
  WeakVH &BBEntry = VMap[BB];

  // Have we already cloned this block?
  if (BBEntry) return;
  
  // Nope, clone it now.
  BasicBlock *NewBB;
  BBEntry = NewBB = BasicBlock::Create(BB->getContext());
  if (BB->hasName()) NewBB->setName(BB->getName()+NameSuffix);

  // It is only legal to clone a function if a block address within that
  // function is never referenced outside of the function.  Given that, we
  // want to map block addresses from the old function to block addresses in
  // the clone. (This is different from the generic ValueMapper
  // implementation, which generates an invalid blockaddress when
  // cloning a function.)
  //
  // Note that we don't need to fix the mapping for unreachable blocks;
  // the default mapping there is safe.
  if (BB->hasAddressTaken()) {
    Constant *OldBBAddr = BlockAddress::get(const_cast<Function*>(OldFunc),
                                            const_cast<BasicBlock*>(BB));
    VMap[OldBBAddr] = BlockAddress::get(NewFunc, NewBB);
  }
    

  bool hasCalls = false, hasDynamicAllocas = false, hasStaticAllocas = false;
  
  // Loop over all instructions, and copy them over, DCE'ing as we go.  This
  // loop doesn't include the terminator.
  for (BasicBlock::const_iterator II = BB->begin(), IE = --BB->end();
       II != IE; ++II) {
    Instruction *NewInst = II->clone();

    // Eagerly remap operands to the newly cloned instruction, except for PHI
    // nodes for which we defer processing until we update the CFG.
    if (!isa<PHINode>(NewInst)) {
      RemapInstruction(NewInst, VMap,
                       ModuleLevelChanges ? RF_None : RF_NoModuleLevelChanges);

      // If we can simplify this instruction to some other value, simply add
      // a mapping to that value rather than inserting a new instruction into
      // the basic block.
      if (Value *V = SimplifyInstruction(NewInst, TD)) {
        // On the off-chance that this simplifies to an instruction in the old
        // function, map it back into the new function.
        if (Value *MappedV = VMap.lookup(V))
          V = MappedV;

        VMap[II] = V;
        delete NewInst;
        continue;
      }
    }

    if (II->hasName())
      NewInst->setName(II->getName()+NameSuffix);
    VMap[II] = NewInst;                // Add instruction map to value.
    NewBB->getInstList().push_back(NewInst);
    hasCalls |= (isa<CallInst>(II) && !isa<DbgInfoIntrinsic>(II));
    if (const AllocaInst *AI = dyn_cast<AllocaInst>(II)) {
      if (isa<ConstantInt>(AI->getArraySize()))
        hasStaticAllocas = true;
      else
        hasDynamicAllocas = true;
    }
  }
  
  // Finally, clone over the terminator.
  const TerminatorInst *OldTI = BB->getTerminator();
  bool TerminatorDone = false;
  if (const BranchInst *BI = dyn_cast<BranchInst>(OldTI)) {
    if (BI->isConditional()) {
      // If the condition was a known constant in the callee...
      ConstantInt *Cond = dyn_cast<ConstantInt>(BI->getCondition());
      // Or is a known constant in the caller...
      if (Cond == 0) {
        Value *V = VMap[BI->getCondition()];
        Cond = dyn_cast_or_null<ConstantInt>(V);
      }

      // Constant fold to uncond branch!
      if (Cond) {
        BasicBlock *Dest = BI->getSuccessor(!Cond->getZExtValue());
        VMap[OldTI] = BranchInst::Create(Dest, NewBB);
        ToClone.push_back(Dest);
        TerminatorDone = true;
      }
    }
  } else if (const SwitchInst *SI = dyn_cast<SwitchInst>(OldTI)) {
    // If switching on a value known constant in the caller.
    ConstantInt *Cond = dyn_cast<ConstantInt>(SI->getCondition());
    if (Cond == 0) { // Or known constant after constant prop in the callee...
      Value *V = VMap[SI->getCondition()];
      Cond = dyn_cast_or_null<ConstantInt>(V);
    }
//.........这里部分代码省略.........

// If we can determine that all possible objects pointed to by the provided
// pointer value are, not only dereferenceable, but also definitively less than
// or equal to the provided maximum size, then return true. Otherwise, return
// false (constant global values and allocas fall into this category).
//
// FIXME: This should probably live in ValueTracking (or similar).
static bool isObjectSizeLessThanOrEq(Value *V, uint64_t MaxSize,
                                     const DataLayout &DL) {
  SmallPtrSet<Value *, 4> Visited;
  SmallVector<Value *, 4> Worklist(1, V);

  do {
    Value *P = Worklist.pop_back_val();
    P = P->stripPointerCasts();

    if (!Visited.insert(P).second)
      continue;

    if (SelectInst *SI = dyn_cast<SelectInst>(P)) {
      Worklist.push_back(SI->getTrueValue());
      Worklist.push_back(SI->getFalseValue());
      continue;
    }

    if (PHINode *PN = dyn_cast<PHINode>(P)) {
      for (Value *IncValue : PN->incoming_values())
        Worklist.push_back(IncValue);
      continue;
    }

    if (GlobalAlias *GA = dyn_cast<GlobalAlias>(P)) {
      if (GA->mayBeOverridden())
        return false;
      Worklist.push_back(GA->getAliasee());
      continue;
    }

    // If we know how big this object is, and it is less than MaxSize, continue
    // searching. Otherwise, return false.
    if (AllocaInst *AI = dyn_cast<AllocaInst>(P)) {
      if (!AI->getAllocatedType()->isSized())
        return false;

      ConstantInt *CS = dyn_cast<ConstantInt>(AI->getArraySize());
      if (!CS)
        return false;

      uint64_t TypeSize = DL.getTypeAllocSize(AI->getAllocatedType());
      // Make sure that, even if the multiplication below would wrap as an
      // uint64_t, we still do the right thing.
      if ((CS->getValue().zextOrSelf(128)*APInt(128, TypeSize)).ugt(MaxSize))
        return false;
      continue;
    }

    if (GlobalVariable *GV = dyn_cast<GlobalVariable>(P)) {
      if (!GV->hasDefinitiveInitializer() || !GV->isConstant())
        return false;

      uint64_t InitSize = DL.getTypeAllocSize(GV->getValueType());
      if (InitSize > MaxSize)
        return false;
      continue;
    }

    return false;
  } while (!Worklist.empty());

  return true;
}

static LazyValueInfo::Tristate getPredicateResult(unsigned Pred, Constant *C,
                                                  LVILatticeVal &Result,
                                                  const DataLayout &DL,
                                                  TargetLibraryInfo *TLI) {

  // If we know the value is a constant, evaluate the conditional.
  Constant *Res = nullptr;
  if (Result.isConstant()) {
    Res = ConstantFoldCompareInstOperands(Pred, Result.getConstant(), C, DL,
                                          TLI);
    if (ConstantInt *ResCI = dyn_cast<ConstantInt>(Res))
      return ResCI->isZero() ? LazyValueInfo::False : LazyValueInfo::True;
    return LazyValueInfo::Unknown;
  }

  if (Result.isConstantRange()) {
    ConstantInt *CI = dyn_cast<ConstantInt>(C);
    if (!CI) return LazyValueInfo::Unknown;

    ConstantRange CR = Result.getConstantRange();
    if (Pred == ICmpInst::ICMP_EQ) {
      if (!CR.contains(CI->getValue()))
        return LazyValueInfo::False;

      if (CR.isSingleElement() && CR.contains(CI->getValue()))
        return LazyValueInfo::True;
    } else if (Pred == ICmpInst::ICMP_NE) {
      if (!CR.contains(CI->getValue()))
        return LazyValueInfo::True;

      if (CR.isSingleElement() && CR.contains(CI->getValue()))
        return LazyValueInfo::False;
    }

    // Handle more complex predicates.
    ConstantRange TrueValues =
        ICmpInst::makeConstantRange((ICmpInst::Predicate)Pred, CI->getValue());
    if (TrueValues.contains(CR))
      return LazyValueInfo::True;
    if (TrueValues.inverse().contains(CR))
      return LazyValueInfo::False;
    return LazyValueInfo::Unknown;
  }

  if (Result.isNotConstant()) {
    // If this is an equality comparison, we can try to fold it knowing that
    // "V != C1".
    if (Pred == ICmpInst::ICMP_EQ) {
      // !C1 == C -> false iff C1 == C.
      Res = ConstantFoldCompareInstOperands(ICmpInst::ICMP_NE,
                                            Result.getNotConstant(), C, DL,
                                            TLI);
      if (Res->isNullValue())
        return LazyValueInfo::False;
    } else if (Pred == ICmpInst::ICMP_NE) {
      // !C1 != C -> true iff C1 == C.
      Res = ConstantFoldCompareInstOperands(ICmpInst::ICMP_NE,
                                            Result.getNotConstant(), C, DL,
                                            TLI);
      if (Res->isNullValue())
        return LazyValueInfo::True;
    }
    return LazyValueInfo::Unknown;
  }

  return LazyValueInfo::Unknown;
}

/// processMemCpyMemCpyDependence - We've found that the (upward scanning)
/// memory dependence of memcpy 'M' is the memcpy 'MDep'.  Try to simplify M to
/// copy from MDep's input if we can.  MSize is the size of M's copy.
/// 
bool MemCpyOpt::processMemCpyMemCpyDependence(MemCpyInst *M, MemCpyInst *MDep,
                                              uint64_t MSize) {
  // We can only transforms memcpy's where the dest of one is the source of the
  // other.
  if (M->getSource() != MDep->getDest() || MDep->isVolatile())
    return false;
  
  // If dep instruction is reading from our current input, then it is a noop
  // transfer and substituting the input won't change this instruction.  Just
  // ignore the input and let someone else zap MDep.  This handles cases like:
  //    memcpy(a <- a)
  //    memcpy(b <- a)
  if (M->getSource() == MDep->getSource())
    return false;
  
  // Second, the length of the memcpy's must be the same, or the preceding one
  // must be larger than the following one.
  ConstantInt *MDepLen = dyn_cast<ConstantInt>(MDep->getLength());
  ConstantInt *MLen = dyn_cast<ConstantInt>(M->getLength());
  if (!MDepLen || !MLen || MDepLen->getZExtValue() < MLen->getZExtValue())
    return false;
  
  AliasAnalysis &AA = getAnalysis<AliasAnalysis>();

  // Verify that the copied-from memory doesn't change in between the two
  // transfers.  For example, in:
  //    memcpy(a <- b)
  //    *b = 42;
  //    memcpy(c <- a)
  // It would be invalid to transform the second memcpy into memcpy(c <- b).
  //
  // TODO: If the code between M and MDep is transparent to the destination "c",
  // then we could still perform the xform by moving M up to the first memcpy.
  //
  // NOTE: This is conservative, it will stop on any read from the source loc,
  // not just the defining memcpy.
  MemDepResult SourceDep =
    MD->getPointerDependencyFrom(AA.getLocationForSource(MDep),
                                 false, M, M->getParent());
  if (!SourceDep.isClobber() || SourceDep.getInst() != MDep)
    return false;
  
  // If the dest of the second might alias the source of the first, then the
  // source and dest might overlap.  We still want to eliminate the intermediate
  // value, but we have to generate a memmove instead of memcpy.
  bool UseMemMove = false;
  if (!AA.isNoAlias(AA.getLocationForDest(M), AA.getLocationForSource(MDep)))
    UseMemMove = true;
  
  // If all checks passed, then we can transform M.
  
  // Make sure to use the lesser of the alignment of the source and the dest
  // since we're changing where we're reading from, but don't want to increase
  // the alignment past what can be read from or written to.
  // TODO: Is this worth it if we're creating a less aligned memcpy? For
  // example we could be moving from movaps -> movq on x86.
  unsigned Align = std::min(MDep->getAlignment(), M->getAlignment());
  
  IRBuilder<> Builder(M);
  if (UseMemMove)
    Builder.CreateMemMove(M->getRawDest(), MDep->getRawSource(), M->getLength(),
                          Align, M->isVolatile());
  else
    Builder.CreateMemCpy(M->getRawDest(), MDep->getRawSource(), M->getLength(),
                         Align, M->isVolatile());

  // Remove the instruction we're replacing.
  MD->removeInstruction(M);
  M->eraseFromParent();
  ++NumMemCpyInstr;
  return true;
}

bool InductionDescriptor::isInductionPHI(
    PHINode *Phi, const Loop *TheLoop, ScalarEvolution *SE,
    InductionDescriptor &D, const SCEV *Expr,
    SmallVectorImpl<Instruction *> *CastsToIgnore) {
  Type *PhiTy = Phi->getType();
  // We only handle integer and pointer inductions variables.
  if (!PhiTy->isIntegerTy() && !PhiTy->isPointerTy())
    return false;

  // Check that the PHI is consecutive.
  const SCEV *PhiScev = Expr ? Expr : SE->getSCEV(Phi);
  const SCEVAddRecExpr *AR = dyn_cast<SCEVAddRecExpr>(PhiScev);

  if (!AR) {
    LLVM_DEBUG(dbgs() << "LV: PHI is not a poly recurrence.\n");
    return false;
  }

  if (AR->getLoop() != TheLoop) {
    // FIXME: We should treat this as a uniform. Unfortunately, we
    // don't currently know how to handled uniform PHIs.
    LLVM_DEBUG(
        dbgs() << "LV: PHI is a recurrence with respect to an outer loop.\n");
    return false;
  }

  Value *StartValue =
      Phi->getIncomingValueForBlock(AR->getLoop()->getLoopPreheader());

  BasicBlock *Latch = AR->getLoop()->getLoopLatch();
  if (!Latch)
    return false;
  BinaryOperator *BOp =
      dyn_cast<BinaryOperator>(Phi->getIncomingValueForBlock(Latch));

  const SCEV *Step = AR->getStepRecurrence(*SE);
  // Calculate the pointer stride and check if it is consecutive.
  // The stride may be a constant or a loop invariant integer value.
  const SCEVConstant *ConstStep = dyn_cast<SCEVConstant>(Step);
  if (!ConstStep && !SE->isLoopInvariant(Step, TheLoop))
    return false;

  if (PhiTy->isIntegerTy()) {
    D = InductionDescriptor(StartValue, IK_IntInduction, Step, BOp,
                            CastsToIgnore);
    return true;
  }

  assert(PhiTy->isPointerTy() && "The PHI must be a pointer");
  // Pointer induction should be a constant.
  if (!ConstStep)
    return false;

  ConstantInt *CV = ConstStep->getValue();
  Type *PointerElementType = PhiTy->getPointerElementType();
  // The pointer stride cannot be determined if the pointer element type is not
  // sized.
  if (!PointerElementType->isSized())
    return false;

  const DataLayout &DL = Phi->getModule()->getDataLayout();
  int64_t Size = static_cast<int64_t>(DL.getTypeAllocSize(PointerElementType));
  if (!Size)
    return false;

  int64_t CVSize = CV->getSExtValue();
  if (CVSize % Size)
    return false;
  auto *StepValue =
      SE->getConstant(CV->getType(), CVSize / Size, true /* signed */);
  D = InductionDescriptor(StartValue, IK_PtrInduction, StepValue, BOp);
  return true;
}

SizeOffsetType ObjectSizeOffsetVisitor::visitCallSite(CallSite CS) {
  Optional<AllocFnsTy> FnData =
      getAllocationData(CS.getInstruction(), AnyAlloc, TLI);
  if (!FnData)
    return unknown();

  // handle strdup-like functions separately
  if (FnData->AllocTy == StrDupLike) {
    APInt Size(IntTyBits, GetStringLength(CS.getArgument(0)));
    if (!Size)
      return unknown();

    // strndup limits strlen
    if (FnData->FstParam > 0) {
      ConstantInt *Arg =
          dyn_cast<ConstantInt>(CS.getArgument(FnData->FstParam));
      if (!Arg)
        return unknown();

      APInt MaxSize = Arg->getValue().zextOrSelf(IntTyBits);
      if (Size.ugt(MaxSize))
        Size = MaxSize + 1;
    }
    return std::make_pair(Size, Zero);
  }

  ConstantInt *Arg = dyn_cast<ConstantInt>(CS.getArgument(FnData->FstParam));
  if (!Arg)
    return unknown();

  // When we're compiling N-bit code, and the user uses parameters that are
  // greater than N bits (e.g. uint64_t on a 32-bit build), we can run into
  // trouble with APInt size issues. This function handles resizing + overflow
  // checks for us.
  auto CheckedZextOrTrunc = [&](APInt &I) {
    // More bits than we can handle. Checking the bit width isn't necessary, but
    // it's faster than checking active bits, and should give `false` in the
    // vast majority of cases.
    if (I.getBitWidth() > IntTyBits && I.getActiveBits() > IntTyBits)
      return false;
    if (I.getBitWidth() != IntTyBits)
      I = I.zextOrTrunc(IntTyBits);
    return true;
  };

  APInt Size = Arg->getValue();
  if (!CheckedZextOrTrunc(Size))
    return unknown();

  // size determined by just 1 parameter
  if (FnData->SndParam < 0)
    return std::make_pair(Size, Zero);

  Arg = dyn_cast<ConstantInt>(CS.getArgument(FnData->SndParam));
  if (!Arg)
    return unknown();

  APInt NumElems = Arg->getValue();
  if (!CheckedZextOrTrunc(NumElems))
    return unknown();

  bool Overflow;
  Size = Size.umul_ov(NumElems, Overflow);
  return Overflow ? unknown() : std::make_pair(Size, Zero);

  // TODO: handle more standard functions (+ wchar cousins):
  // - strdup / strndup
  // - strcpy / strncpy
  // - strcat / strncat
  // - memcpy / memmove
  // - strcat / strncat
  // - memset
}

int qdp_jit_vec::vectorize_loads( std::vector<std::vector<Instruction*> >& load_instructions )
{
  DEBUG(dbgs() << "Vectorize loads, total of " << load_instructions.size() << "\n");

  //std::vector<std::pair<Value*,Value*> > scalar_vector_loads;
  scalar_vector_pairs.clear();

  if (load_instructions.empty())
    return 0;

  int load_vec_elem = 0;
  for( std::vector<Instruction*>& VI : load_instructions ) {
    DEBUG(dbgs() << "Processing vector of loads number " << load_vec_elem++ << "\n");
    assert( VI.size() == vec_len && "length of vector of loads does not match vec_len" );
    int loads_consec = true;
    uint64_t lo,hi;
    bool first = true;
    for( Instruction* I : VI ) {
      GetElementPtrInst* GEP;
      if ((GEP = dyn_cast<GetElementPtrInst>(I->getOperand(0)))) {
	if (first) {
	  ConstantInt * CI;
	  if ((CI = dyn_cast<ConstantInt>(GEP->getOperand(1)))) {
	    lo = CI->getZExtValue();
	    hi = lo+1;
	    first=false;
	  } else {
	    DEBUG(dbgs() << "First load in the chain: Operand of GEP not a ConstantInt" << *GEP->getOperand(1) << "\n");
	    assert( 0 && "First load in the chain: Operand of GEP not a ConstantInt\n");
	    exit(0);
	  }
	} else {
	  ConstantInt * CI;
	  if ((CI = dyn_cast<ConstantInt>(GEP->getOperand(1)))) {
	    if (hi != CI->getZExtValue()) {
	      DEBUG(dbgs() << "Loads not consecutive lo=" << lo << " hi=" << hi << " this=" << CI->getZExtValue() << "\n");
	      loads_consec = false;
	    } else {
	      hi++;
	    }
	  }
	}
      } else {
	DEBUG(dbgs() << "Operand of load not a GEP " << *I->getOperand(0) << "\n");
	assert( 0 && "Operand of load not a GEP" );
	exit(0);
	loads_consec = false;
      }
    }
    if (loads_consec) {
      DEBUG(dbgs() << "Loads consecuetive\n");

      LoadInst* LI = cast<LoadInst>(VI.at(0));
      GetElementPtrInst* GEP = cast<GetElementPtrInst>(LI->getOperand(0));
      Instruction* GEPcl = clone_with_operands(GEP);
      unsigned AS = LI->getPointerAddressSpace();
      VectorType *VecTy = VectorType::get( LI->getType() , vec_len );

      unsigned bitwidth = LI->getType()->getPrimitiveSizeInBits();
      unsigned bytewidth = bitwidth == 1 ? 1 : bitwidth/8;
      DEBUG(dbgs() << "bit/byte width of load instr trype: " << bitwidth << "/" << bytewidth << "\n");
 
      //Builder->SetInsertPoint( GEP );
      Value *VecPtr = Builder->CreateBitCast(GEPcl,VecTy->getPointerTo(AS));
      //Value *VecLoad = Builder->CreateLoad( VecPtr );
      
      unsigned align = lo % vec_len == 0 ? bytewidth * vec_len : bytewidth;
      Value *VecLoad = Builder->CreateAlignedLoad( VecPtr , align );

      //DEBUG(dbgs() << "created vector load: " << *VecLoad << "\n");
      //function->dump();

      // unsigned AS = LI->getPointerAddressSpace();
      // VectorType *VecTy = VectorType::get( LI->getType() , vec_len );
      // Builder->SetInsertPoint( LI );
      // Value *VecPtr = Builder->CreateBitCast(LI->getPointerOperand(),VecTy->getPointerTo(AS));
      // Value *VecLoad = Builder->CreateLoad( VecPtr );

      scalar_vector_pairs.push_back( std::make_pair( VI.at(0) , VecLoad ) );
    } else {
      DEBUG(dbgs() << "Loads not consecutive:\n");
      for (Value* V: VI) {
	DEBUG(dbgs() << *V << "\n");
      }

      //Instruction* I = dyn_cast<Instruction>(VI.back()->getNextNode());
      //DEBUG(dbgs() << *I << "\n");

      //Builder->SetInsertPoint( VI.at(0) );


      std::vector<Instruction*> VIcl;
      for( Instruction* I : VI ) {
	VIcl.push_back( clone_with_operands(I) );
      }

      VectorType *VecTy = VectorType::get( VI.at(0)->getType() , vec_len );
      Value *Vec = UndefValue::get(VecTy);

      int i=0;
//.........这里部分代码省略.........

/// CanEvaluateShifted - See if we can compute the specified value, but shifted
/// logically to the left or right by some number of bits.  This should return
/// true if the expression can be computed for the same cost as the current
/// expression tree.  This is used to eliminate extraneous shifting from things
/// like:
///      %C = shl i128 %A, 64
///      %D = shl i128 %B, 96
///      %E = or i128 %C, %D
///      %F = lshr i128 %E, 64
/// where the client will ask if E can be computed shifted right by 64-bits.  If
/// this succeeds, the GetShiftedValue function will be called to produce the
/// value.
static bool CanEvaluateShifted(Value *V, unsigned NumBits, bool isLeftShift,
                               InstCombiner &IC) {
  // We can always evaluate constants shifted.
  if (isa<Constant>(V))
    return true;

  Instruction *I = dyn_cast<Instruction>(V);
  if (!I) return false;

  // If this is the opposite shift, we can directly reuse the input of the shift
  // if the needed bits are already zero in the input.  This allows us to reuse
  // the value which means that we don't care if the shift has multiple uses.
  //  TODO:  Handle opposite shift by exact value.
  ConstantInt *CI = 0;
  if ((isLeftShift && match(I, m_LShr(m_Value(), m_ConstantInt(CI)))) ||
      (!isLeftShift && match(I, m_Shl(m_Value(), m_ConstantInt(CI))))) {
    if (CI->getZExtValue() == NumBits) {
      // TODO: Check that the input bits are already zero with MaskedValueIsZero
#if 0
      // If this is a truncate of a logical shr, we can truncate it to a smaller
      // lshr iff we know that the bits we would otherwise be shifting in are
      // already zeros.
      uint32_t OrigBitWidth = OrigTy->getScalarSizeInBits();
      uint32_t BitWidth = Ty->getScalarSizeInBits();
      if (MaskedValueIsZero(I->getOperand(0),
            APInt::getHighBitsSet(OrigBitWidth, OrigBitWidth-BitWidth)) &&
          CI->getLimitedValue(BitWidth) < BitWidth) {
        return CanEvaluateTruncated(I->getOperand(0), Ty);
      }
#endif

    }
  }

  // We can't mutate something that has multiple uses: doing so would
  // require duplicating the instruction in general, which isn't profitable.
  if (!I->hasOneUse()) return false;

  switch (I->getOpcode()) {
  default: return false;
  case Instruction::And:
  case Instruction::Or:
  case Instruction::Xor:
    // Bitwise operators can all arbitrarily be arbitrarily evaluated shifted.
    return CanEvaluateShifted(I->getOperand(0), NumBits, isLeftShift, IC) &&
           CanEvaluateShifted(I->getOperand(1), NumBits, isLeftShift, IC);

  case Instruction::Shl: {
    // We can often fold the shift into shifts-by-a-constant.
    CI = dyn_cast<ConstantInt>(I->getOperand(1));
    if (CI == 0) return false;

    // We can always fold shl(c1)+shl(c2) -> shl(c1+c2).
    if (isLeftShift) return true;

    // We can always turn shl(c)+shr(c) -> and(c2).
    if (CI->getValue() == NumBits) return true;

    unsigned TypeWidth = I->getType()->getScalarSizeInBits();

    // We can turn shl(c1)+shr(c2) -> shl(c3)+and(c4), but it isn't
    // profitable unless we know the and'd out bits are already zero.
    if (CI->getZExtValue() > NumBits) {
      unsigned LowBits = TypeWidth - CI->getZExtValue();
      if (MaskedValueIsZero(I->getOperand(0),
                       APInt::getLowBitsSet(TypeWidth, NumBits) << LowBits))
        return true;
    }

    return false;
  }
  case Instruction::LShr: {
    // We can often fold the shift into shifts-by-a-constant.
    CI = dyn_cast<ConstantInt>(I->getOperand(1));
    if (CI == 0) return false;

    // We can always fold lshr(c1)+lshr(c2) -> lshr(c1+c2).
    if (!isLeftShift) return true;

    // We can always turn lshr(c)+shl(c) -> and(c2).
    if (CI->getValue() == NumBits) return true;

    unsigned TypeWidth = I->getType()->getScalarSizeInBits();

    // We can always turn lshr(c1)+shl(c2) -> lshr(c3)+and(c4), but it isn't
    // profitable unless we know the and'd out bits are already zero.
    if (CI->getValue().ult(TypeWidth) && CI->getZExtValue() > NumBits) {
      unsigned LowBits = CI->getZExtValue() - NumBits;
//.........这里部分代码省略.........

bool FastISel::SelectCall(const User *I) {
  const CallInst *Call = cast<CallInst>(I);

  // Handle simple inline asms.
  if (const InlineAsm *IA = dyn_cast<InlineAsm>(Call->getCalledValue())) {
    // Don't attempt to handle constraints.
    if (!IA->getConstraintString().empty())
      return false;

    unsigned ExtraInfo = 0;
    if (IA->hasSideEffects())
      ExtraInfo |= InlineAsm::Extra_HasSideEffects;
    if (IA->isAlignStack())
      ExtraInfo |= InlineAsm::Extra_IsAlignStack;

    BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
            TII.get(TargetOpcode::INLINEASM))
      .addExternalSymbol(IA->getAsmString().c_str())
      .addImm(ExtraInfo);
    return true;
  }

  MachineModuleInfo &MMI = FuncInfo.MF->getMMI();
  ComputeUsesVAFloatArgument(*Call, &MMI);

  const Function *F = Call->getCalledFunction();
  if (!F) return false;

  // Handle selected intrinsic function calls.
  switch (F->getIntrinsicID()) {
  default: break;
    // At -O0 we don't care about the lifetime intrinsics.
  case Intrinsic::lifetime_start:
  case Intrinsic::lifetime_end:
    return true;
  case Intrinsic::dbg_declare: {
    const DbgDeclareInst *DI = cast<DbgDeclareInst>(Call);
    if (!DIVariable(DI->getVariable()).Verify() ||
        !FuncInfo.MF->getMMI().hasDebugInfo()) {
      DEBUG(dbgs() << "Dropping debug info for " << *DI << "\n");
      return true;
    }

    const Value *Address = DI->getAddress();
    if (!Address || isa<UndefValue>(Address)) {
      DEBUG(dbgs() << "Dropping debug info for " << *DI << "\n");
      return true;
    }

    unsigned Reg = 0;
    unsigned Offset = 0;
    if (const Argument *Arg = dyn_cast<Argument>(Address)) {
      // Some arguments' frame index is recorded during argument lowering.
      Offset = FuncInfo.getArgumentFrameIndex(Arg);
      if (Offset)
        Reg = TRI.getFrameRegister(*FuncInfo.MF);
    }
    if (!Reg)
      Reg = lookUpRegForValue(Address);

    if (!Reg && isa<Instruction>(Address) &&
        (!isa<AllocaInst>(Address) ||
         !FuncInfo.StaticAllocaMap.count(cast<AllocaInst>(Address))))
      Reg = FuncInfo.InitializeRegForValue(Address);

    if (Reg)
      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL,
              TII.get(TargetOpcode::DBG_VALUE))
        .addReg(Reg, RegState::Debug).addImm(Offset)
        .addMetadata(DI->getVariable());
    else
      // We can't yet handle anything else here because it would require
      // generating code, thus altering codegen because of debug info.
      DEBUG(dbgs() << "Dropping debug info for " << DI);
    return true;
  }
  case Intrinsic::dbg_value: {
    // This form of DBG_VALUE is target-independent.
    const DbgValueInst *DI = cast<DbgValueInst>(Call);
    const MCInstrDesc &II = TII.get(TargetOpcode::DBG_VALUE);
    const Value *V = DI->getValue();
    if (!V) {
      // Currently the optimizer can produce this; insert an undef to
      // help debugging.  Probably the optimizer should not do this.
      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
        .addReg(0U).addImm(DI->getOffset())
        .addMetadata(DI->getVariable());
    } else if (const ConstantInt *CI = dyn_cast<ConstantInt>(V)) {
      if (CI->getBitWidth() > 64)
        BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
          .addCImm(CI).addImm(DI->getOffset())
          .addMetadata(DI->getVariable());
      else 
        BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
          .addImm(CI->getZExtValue()).addImm(DI->getOffset())
          .addMetadata(DI->getVariable());
    } else if (const ConstantFP *CF = dyn_cast<ConstantFP>(V)) {
      BuildMI(*FuncInfo.MBB, FuncInfo.InsertPt, DL, II)
        .addFPImm(CF).addImm(DI->getOffset())
        .addMetadata(DI->getVariable());
//.........这里部分代码省略.........

/// GetShiftedValue - When CanEvaluateShifted returned true for an expression,
/// this value inserts the new computation that produces the shifted value.
static Value *GetShiftedValue(Value *V, unsigned NumBits, bool isLeftShift,
                              InstCombiner &IC) {
  // We can always evaluate constants shifted.
  if (Constant *C = dyn_cast<Constant>(V)) {
    if (isLeftShift)
      V = IC.Builder->CreateShl(C, NumBits);
    else
      V = IC.Builder->CreateLShr(C, NumBits);
    // If we got a constantexpr back, try to simplify it with TD info.
    if (ConstantExpr *CE = dyn_cast<ConstantExpr>(V))
      V = ConstantFoldConstantExpression(CE, IC.getDataLayout(),
                                         IC.getTargetLibraryInfo());
    return V;
  }

  Instruction *I = cast<Instruction>(V);
  IC.Worklist.Add(I);

  switch (I->getOpcode()) {
  default: llvm_unreachable("Inconsistency with CanEvaluateShifted");
  case Instruction::And:
  case Instruction::Or:
  case Instruction::Xor:
    // Bitwise operators can all arbitrarily be arbitrarily evaluated shifted.
    I->setOperand(0, GetShiftedValue(I->getOperand(0), NumBits,isLeftShift,IC));
    I->setOperand(1, GetShiftedValue(I->getOperand(1), NumBits,isLeftShift,IC));
    return I;

  case Instruction::Shl: {
    BinaryOperator *BO = cast<BinaryOperator>(I);
    unsigned TypeWidth = BO->getType()->getScalarSizeInBits();

    // We only accept shifts-by-a-constant in CanEvaluateShifted.
    ConstantInt *CI = cast<ConstantInt>(BO->getOperand(1));

    // We can always fold shl(c1)+shl(c2) -> shl(c1+c2).
    if (isLeftShift) {
      // If this is oversized composite shift, then unsigned shifts get 0.
      unsigned NewShAmt = NumBits+CI->getZExtValue();
      if (NewShAmt >= TypeWidth)
        return Constant::getNullValue(I->getType());

      BO->setOperand(1, ConstantInt::get(BO->getType(), NewShAmt));
      BO->setHasNoUnsignedWrap(false);
      BO->setHasNoSignedWrap(false);
      return I;
    }

    // We turn shl(c)+lshr(c) -> and(c2) if the input doesn't already have
    // zeros.
    if (CI->getValue() == NumBits) {
      APInt Mask(APInt::getLowBitsSet(TypeWidth, TypeWidth - NumBits));
      V = IC.Builder->CreateAnd(BO->getOperand(0),
                                ConstantInt::get(BO->getContext(), Mask));
      if (Instruction *VI = dyn_cast<Instruction>(V)) {
        VI->moveBefore(BO);
        VI->takeName(BO);
      }
      return V;
    }

    // We turn shl(c1)+shr(c2) -> shl(c3)+and(c4), but only when we know that
    // the and won't be needed.
    assert(CI->getZExtValue() > NumBits);
    BO->setOperand(1, ConstantInt::get(BO->getType(),
                                       CI->getZExtValue() - NumBits));
    BO->setHasNoUnsignedWrap(false);
    BO->setHasNoSignedWrap(false);
    return BO;
  }
  case Instruction::LShr: {
    BinaryOperator *BO = cast<BinaryOperator>(I);
    unsigned TypeWidth = BO->getType()->getScalarSizeInBits();
    // We only accept shifts-by-a-constant in CanEvaluateShifted.
    ConstantInt *CI = cast<ConstantInt>(BO->getOperand(1));

    // We can always fold lshr(c1)+lshr(c2) -> lshr(c1+c2).
    if (!isLeftShift) {
      // If this is oversized composite shift, then unsigned shifts get 0.
      unsigned NewShAmt = NumBits+CI->getZExtValue();
      if (NewShAmt >= TypeWidth)
        return Constant::getNullValue(BO->getType());

      BO->setOperand(1, ConstantInt::get(BO->getType(), NewShAmt));
      BO->setIsExact(false);
      return I;
    }

    // We turn lshr(c)+shl(c) -> and(c2) if the input doesn't already have
    // zeros.
    if (CI->getValue() == NumBits) {
      APInt Mask(APInt::getHighBitsSet(TypeWidth, TypeWidth - NumBits));
      V = IC.Builder->CreateAnd(I->getOperand(0),
                                ConstantInt::get(BO->getContext(), Mask));
      if (Instruction *VI = dyn_cast<Instruction>(V)) {
        VI->moveBefore(I);
        VI->takeName(I);
      }
//.........这里部分代码省略.........

/// Return true if we can evaluate the specified expression tree if the vector
/// elements were shuffled in a different order.
static bool CanEvaluateShuffled(Value *V, ArrayRef<int> Mask,
                                unsigned Depth = 5) {
  // We can always reorder the elements of a constant.
  if (isa<Constant>(V))
    return true;

  // We won't reorder vector arguments. No IPO here.
  Instruction *I = dyn_cast<Instruction>(V);
  if (!I) return false;

  // Two users may expect different orders of the elements. Don't try it.
  if (!I->hasOneUse())
    return false;

  if (Depth == 0) return false;

  switch (I->getOpcode()) {
    case Instruction::Add:
    case Instruction::FAdd:
    case Instruction::Sub:
    case Instruction::FSub:
    case Instruction::Mul:
    case Instruction::FMul:
    case Instruction::UDiv:
    case Instruction::SDiv:
    case Instruction::FDiv:
    case Instruction::URem:
    case Instruction::SRem:
    case Instruction::FRem:
    case Instruction::Shl:
    case Instruction::LShr:
    case Instruction::AShr:
    case Instruction::And:
    case Instruction::Or:
    case Instruction::Xor:
    case Instruction::ICmp:
    case Instruction::FCmp:
    case Instruction::Trunc:
    case Instruction::ZExt:
    case Instruction::SExt:
    case Instruction::FPToUI:
    case Instruction::FPToSI:
    case Instruction::UIToFP:
    case Instruction::SIToFP:
    case Instruction::FPTrunc:
    case Instruction::FPExt:
    case Instruction::GetElementPtr: {
      for (Value *Operand : I->operands()) {
        if (!CanEvaluateShuffled(Operand, Mask, Depth-1))
          return false;
      }
      return true;
    }
    case Instruction::InsertElement: {
      ConstantInt *CI = dyn_cast<ConstantInt>(I->getOperand(2));
      if (!CI) return false;
      int ElementNumber = CI->getLimitedValue();

      // Verify that 'CI' does not occur twice in Mask. A single 'insertelement'
      // can't put an element into multiple indices.
      bool SeenOnce = false;
      for (int i = 0, e = Mask.size(); i != e; ++i) {
        if (Mask[i] == ElementNumber) {
          if (SeenOnce)
            return false;
          SeenOnce = true;
        }
      }
      return CanEvaluateShuffled(I->getOperand(0), Mask, Depth-1);
    }
  }
  return false;
}

/// \brief Check if Value is always a dereferenceable pointer.
///
/// Test if V is always a pointer to allocated and suitably aligned memory for
/// a simple load or store.
static bool isDereferenceablePointer(const Value *V, const DataLayout *DL,
                                     SmallPtrSetImpl<const Value *> &Visited) {
  // Note that it is not safe to speculate into a malloc'd region because
  // malloc may return null.

  // These are obviously ok.
  if (isa<AllocaInst>(V)) return true;

  // It's not always safe to follow a bitcast, for example:
  //   bitcast i8* (alloca i8) to i32*
  // would result in a 4-byte load from a 1-byte alloca. However,
  // if we're casting from a pointer from a type of larger size
  // to a type of smaller size (or the same size), and the alignment
  // is at least as large as for the resulting pointer type, then
  // we can look through the bitcast.
  if (DL)
    if (const BitCastInst* BC = dyn_cast<BitCastInst>(V)) {
      Type *STy = BC->getSrcTy()->getPointerElementType(),
           *DTy = BC->getDestTy()->getPointerElementType();
      if (STy->isSized() && DTy->isSized() &&
          (DL->getTypeStoreSize(STy) >=
           DL->getTypeStoreSize(DTy)) &&
          (DL->getABITypeAlignment(STy) >=
           DL->getABITypeAlignment(DTy)))
        return isDereferenceablePointer(BC->getOperand(0), DL, Visited);
    }

  // Global variables which can't collapse to null are ok.
  if (const GlobalVariable *GV = dyn_cast<GlobalVariable>(V))
    return !GV->hasExternalWeakLinkage();

  // byval arguments are okay. Arguments specifically marked as
  // dereferenceable are okay too.
  if (const Argument *A = dyn_cast<Argument>(V)) {
    if (A->hasByValAttr())
      return true;
    else if (uint64_t Bytes = A->getDereferenceableBytes()) {
      Type *Ty = V->getType()->getPointerElementType();
      if (Ty->isSized() && DL && DL->getTypeStoreSize(Ty) <= Bytes)
        return true;
    }

    return false;
  }

  // Return values from call sites specifically marked as dereferenceable are
  // also okay.
  if (ImmutableCallSite CS = V) {
    if (uint64_t Bytes = CS.getDereferenceableBytes(0)) {
      Type *Ty = V->getType()->getPointerElementType();
      if (Ty->isSized() && DL && DL->getTypeStoreSize(Ty) <= Bytes)
        return true;
    }
  }

  // For GEPs, determine if the indexing lands within the allocated object.
  if (const GEPOperator *GEP = dyn_cast<GEPOperator>(V)) {
    // Conservatively require that the base pointer be fully dereferenceable.
    if (!Visited.insert(GEP->getOperand(0)).second)
      return false;
    if (!isDereferenceablePointer(GEP->getOperand(0), DL, Visited))
      return false;
    // Check the indices.
    gep_type_iterator GTI = gep_type_begin(GEP);
    for (User::const_op_iterator I = GEP->op_begin()+1,
         E = GEP->op_end(); I != E; ++I) {
      Value *Index = *I;
      Type *Ty = *GTI++;
      // Struct indices can't be out of bounds.
      if (isa<StructType>(Ty))
        continue;
      ConstantInt *CI = dyn_cast<ConstantInt>(Index);
      if (!CI)
        return false;
      // Zero is always ok.
      if (CI->isZero())
        continue;
      // Check to see that it's within the bounds of an array.
      ArrayType *ATy = dyn_cast<ArrayType>(Ty);
      if (!ATy)
        return false;
      if (CI->getValue().getActiveBits() > 64)
        return false;
      if (CI->getZExtValue() >= ATy->getNumElements())
        return false;
    }
    // Indices check out; this is dereferenceable.
    return true;
  }

  if (const AddrSpaceCastInst *ASC = dyn_cast<AddrSpaceCastInst>(V))
    return isDereferenceablePointer(ASC->getOperand(0), DL, Visited);

  // If we don't know, assume the worst.
  return false;
}

void TartGCPrinter::finishAssembly(AsmPrinter &AP) {
    unsigned nextLabel = 1;
    SafePointList safePoints;

    // Set up for emitting addresses.
    int pointerSize = AP.TM.getTargetData()->getPointerSize();
    int addressAlignLog;
    if (pointerSize == sizeof(int32_t)) {
        addressAlignLog = 2;
    } else {
        addressAlignLog = 3;
    }

    MCStreamer & outStream = AP.OutStreamer;

    // Put this in the data section.
    outStream.SwitchSection(AP.getObjFileLowering().getDataSection());

    // For each function...
    for (iterator FI = begin(), FE = end(); FI != FE; ++FI) {
        GCFunctionInfo & gcFn = **FI;

//    if (optShowGC) {
//      errs() << "GCStrategy: Function: " << gcFn.getFunction().getName() << "\n";
//    }

        // And each safe point...
        for (GCFunctionInfo::iterator sp = gcFn.begin(); sp != gcFn.end(); ++sp) {
            StackTraceTable::FieldOffsetList fieldOffsets;
            StackTraceTable::TraceMethodList traceMethods;

            // And for each live root...
            for (GCFunctionInfo::live_iterator rt = gcFn.live_begin(sp); rt != gcFn.live_end(sp); ++rt) {
                int64_t offset = rt->StackOffset;
                const Constant * meta = rt->Metadata;

                if (meta != NULL && !meta->isNullValue()) {
                    // Meta is non-null, so it's a value type.
                    const ConstantArray * traceArray = cast<ConstantArray>(getGlobalValue(meta));

                    // For each trace descriptor in thre meta array...
                    for (ConstantArray::const_op_iterator it = traceArray->op_begin();
                            it != traceArray->op_end(); ++it) {
                        ConstantStruct * descriptor = cast<ConstantStruct>(*it);
                        ConstantInt * fieldCount = cast<ConstantInt>(descriptor->getOperand(1));
                        int64_t dscOffset = toInt(descriptor->getOperand(2), AP.TM);

                        if (fieldCount->isZero()) {
                            // A zero field count means that this is a trace method descriptor.
                            const Constant * traceMethod = descriptor->getOperand(3);
                            assert(offset > -1000 && offset < 1000);
                            assert(dscOffset > -1000 && dscOffset < 1000);
                            traceMethods.push_back(TraceMethodEntry(offset + dscOffset, traceMethod));
                        } else {
                            // Otherwise it's a field offset descriptor.
                            const GlobalVariable * fieldOffsetsVar = cast<GlobalVariable>(
                                        descriptor->getOperand(3)->getOperand(0));

                            // Handle case where the array value is just a ConstantAggregateZero, which
                            // can be generated by llvm::ConstantArray::get() if the array values
                            // are all zero.
                            if (const ConstantAggregateZero * zero =
                                        dyn_cast<ConstantAggregateZero>(fieldOffsetsVar->getInitializer())) {
                                // Array should never contain duplicate offsets, so an all-zero array
                                // can only have one entry.
                                (void)zero;
                                assert(fieldCount->isOne());
                                fieldOffsets.push_back(offset + dscOffset);
                            } else {
                                // Get the field offset array and add to field offsets for this
                                // safe point.
                                const ConstantArray * fieldOffsetArray = cast<ConstantArray>(
                                            fieldOffsetsVar->getInitializer());
                                for (ConstantArray::const_op_iterator el = fieldOffsetArray->op_begin();
                                        el != fieldOffsetArray->op_end(); ++el) {
                                    fieldOffsets.push_back(
                                        offset + dscOffset + toInt(cast<llvm::Constant>(*el), AP.TM));
                                }
                            }
                        }
                    }
                } else {
                    // No metadata, so it's an object reference - just add the field offset.
                    fieldOffsets.push_back(offset);
                }
            }

            // Nothing to trace? Then we're done.
            if (fieldOffsets.empty() && traceMethods.empty()) {
                continue;
            }

            // Create a folding set node and merge with any identical trace tables.
            std::sort(fieldOffsets.begin(), fieldOffsets.end());
            llvm::FoldingSetNodeID id;
            StackTraceTable::ProfileEntries(id, fieldOffsets, traceMethods);

            void * insertPos;
            StackTraceTable * sTable = traceTables.FindNodeOrInsertPos(id, insertPos);
            if (sTable == NULL) {
//.........这里部分代码省略.........